[PATCH] arm/crash_dump.c: fix Oops caused by overflow converted from pfn to phys

ivan lam ivan.lam2014 at gmail.com
Sun Dec 1 03:32:09 EST 2013

In Cortex-A9 machine, with LPAE feature enabled and physical address space
larger than 4G, Capture kernel will Oops on reading the /proc/vmcore file.

In copy_oldmem_page function, it simply use "pfn << PAGE_SHIFT" to get
the physical address, but this will cause overflow when LPAE enabled and
physical address space is larger than 4G, thus get the result as 0 and
cause Oops.

Fix it to use __pfn_to_phys instead. __pfn_to_phys will first cast the pfn
to type phys_addr_t which is the type u64 when LPAE enabled, and left shifted
by PAGE_SHIFT to get the proper physical address.

Signed-off-by: ivan lam <ivan.lam2014 at gmail.com>
 arch/arm/kernel/crash_dump.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm/kernel/crash_dump.c b/arch/arm/kernel/crash_dump.c
index 90c50d4..5d1286d 100644
--- a/arch/arm/kernel/crash_dump.c
+++ b/arch/arm/kernel/crash_dump.c
@@ -39,7 +39,7 @@ ssize_t copy_oldmem_page(unsigned long pfn, char *buf,
 	if (!csize)
 		return 0;
-	vaddr = ioremap(pfn << PAGE_SHIFT, PAGE_SIZE);
+	vaddr = ioremap(__pfn_to_phys(pfn), PAGE_SIZE);
 	if (!vaddr)
 		return -ENOMEM;

More information about the linux-arm-kernel mailing list